Our understanding of how humans respond to HIV has been revolutionized by the introduction of techniques for isolating anti-viral antibodies from single cells1. Such methods have led to the discovery of naturally occurring, potent antibodies that can neutralize a broad range of HIV viruses, and prevent2 and suppress3 infection. These findings, combined with the association between antibody responses and protection from infection that was identified in a human trial4 of the vaccine RV144, have re-invigorated the quest for antibody-based HIV vaccines. However, it has also become clear that anti-HIV antibodies undergo unusually high levels of mutation1,5, which represents a potential stumbling block for vaccine development. Among four recent studies6,7,8,9 that address this subject is a paper in this issue, in which Liao et al.9 (page 469) track antibody and viral evolution during one patient's response to HIVFootnote 1.

Antibodies are produced by the B lymphocytes of the immune system. The receptors on the surface of each circulating B cell are unique, enabling an immune response to any foreign structure. When a B cell encounters an entity that matches its receptor, it is stimulated to proliferate and secrete antibodies against that structure. Although B-cell genes frequently undergo somatic (non-germline) mutation to increase the affinity of the antibodies they produce, anti-HIV antibodies are unusual in that they are highly somatically mutated — they are therefore quite different from those encoded by the B cells that initially respond to the infection1,10. Furthermore, these mutations seem to be required for the antibodies to bind to heterologous viral-envelope proteins (those expressed on most HIV viruses)5,10. If B cells that express the germline antibody precursor do not bind to the antigen, how are they stimulated in the first place, and why do the antibodies need so many mutations? Answering these questions is of fundamental importance in attempts to reproduce this antibody-development process by vaccination.

Some patients with HIV develop broadly neutralizing antibody activity, but only 2–4 years after infection. Scrutiny of the antibodies produced by single human B cells1 showed that these broadly neutralizing responses are due to a combination of antibodies in some individuals, and to single, potent antibodies in others2. In an attempt to dissect the natural pathways that lead to the generation of these antibodies, Liao et al. studied a patient who developed broad and potent antibodies.

The authors investigated the co-evolution of the HIV-1 virus and the broadly neutralizing antibodies for 34 months from the start of the infection. They isolated a virus-specific antibody named CH103, and clonal variants of it, from single memory B cells that were obtained using a fluorescently tagged viral-envelope protein as bait3,5,11. CH103 neutralizes 55% of HIV-1 isolates and targets the site on the virus that binds to CD4 molecules on the surface of T cells (the immune cells that HIV infects). Like other antibodies in this class5,11, CH103 is highly somatically mutated, and its unmutated germline precursor fails to bind to heterologous HIV-1 envelope proteins9.

One of Liao and colleagues' key findings is that the germline precursor antibody of CH103 has high affinity for the envelope protein expressed by the founder virus that infected the individual. The authors suggest that a progenitor B cell that expresses this germline antibody might only be stimulated to respond if it is presented with the envelope proteins of the founder virus, or similar proteins. The idea that certain envelope proteins are more likely to induce broadly neutralizing antibodies is supported by experiments in macaques showing that specific envelopes induce such responses to simian HIV, whereas others do not12.

However, simply initiating the antibody response is not sufficient for effective immune defence. It takes time and unusually large numbers of somatic mutations for antibody breadth and potency to develop. Liao et al. reconstructed the CH103 clonal lineage by using samples that went back to the time of infection. Although all members of the lineage recognized and neutralized the founder virus, the affinity and neutralizing activity against heterologous viruses gradually increased through the accumulation of somatic mutations. The authors also found that, as previously described for glycan-dependent broadly neutralizing antibodies13, viral diversification and the emergence of 'escape mutants' (those with mutations in the site targeted by the antibody) preceded the development of antibody breadth. By studying a crystal structure of the CH103 antibody in complex with its envelope protein target, Liao et al. showed that HIV escapes antibody pressure by mutating amino-acid residues in and around the CD4 binding site. These resistant viruses then elicit further somatic mutation and 'affinity maturation' of CH103 antibody variants, resulting in greater neutralization breadth of the antibody response.

The reason for the high level of somatic mutation required to produce broadly acting, potent anti-HIV antibodies has recently been investigated6. Under normal circumstances, high affinity of an antibody for its target is usually achieved after the accumulation of 10–15 mutations in the complementarity-determining region of the antibody that forms the antigen contact site. However, broad and potent anti-HIV antibodies contain 40–100 somatic mutations1,5,6,11 that span both the complementarity-determining region and the relatively constant, and mutation-resistant, framework regions. Experiments in which mutations in the framework regions were selectively reverted showed that these mutations are necessary for the evolution of broad and potent anti-HIV antibodies6. These structural alterations in the antibody were found to contribute to direct contacts with the virus and to enhanced flexibility of the antibody structure, both of which are required for optimal breadth and potency.

These data suggest a molecular explanation for why broadly neutralizing anti-HIV antibodies take 2–4 years to develop.

Combined with Liao and colleagues' findings, these data suggest a molecular explanation for why broadly neutralizing anti-HIV antibodies take 2–4 years to develop. Moreover, they indicate that an effective vaccine may require shepherding of B-cell responses through multiple rounds of the natural antibody maturation and mutation process, using naturally derived viral envelopes that induce the production of broad and potent antibodies in people with HIV. A recently suggested7,8alternative, non-mutually exclusive approach is to design specific 'immunogen' molecules that would bind to and activate B cells that produce the germline precursors of broadly neutralizing antibodies. Whether such roadmaps can be used to design effective vaccine strategies has yet to be determined, but they present a strong and testable route to addressing the main challenges of creating an antibody-based HIV-1 vaccine.